Microservices with Consul and Nomad

Transcript

1. mindloops Richard Kettelerij Microservices with
 Consul & Nomad @rkettelerij
2. Intro
3. Agenda • Intro • Use case @ Police • Hashicorp stack • Consul (incl. demo) • Nomad (incl. demo) • Recap
4. Product line Cloud, Big data & Internet
6. Already in place…
7. What could be improved…. • Limited automation around creating new VM’s • No solution for high availability • No zero-downtime deployments - Purpose-built single instance VM for each microservice -
8. Architecture (over a year ago) service-foo service-bar service-qux
9. More VM’s service-foo service-bar service-qux service-foo service-qux service-bar
13. Provisioning tools
14. Runtime tools
15. Suite
16. Consul • Service discovery (via HTTP and DNS) • Health checking • DNS (A record, SRV record) • HTTP • Files: Consul-Template • Environment variables: EnvConsul • SDKs/Frameworks: Spring Cloud Consul, Dropwizard, etc, etc
17. Consul Consul Connect
 (service segmentation) • Service configuration
 (K/V store with watches)
18. service-foo service-bar service-qux service-foo service-qux service-bar More VM’s
19. service-foo service-bar service-qux service-foo service-qux Consul cluster service-bar Service discovery
20. Multiple datacenters test prod OpenStack tenant A test prod OpenStack tenant B service- service- service- service-foo service-qux Consul cluster service-bar Webserver service- service- service- service-foo service-qux Consul cluster service-bar Webserver service- service- service- service-foo service-qux Consul cluster service-bar Webserver service- service- service- service-foo service-qux Consul cluster service-bar Webserver
21. Science • Raft • SWIM
22. Demo time!
24. What is Nomad? Scheduler 
 a.k.a Cluster manager a.k.a Orchestrator
25. What is Nomad?
26. Type of jobs • System • Service • Batch
27. Why Nomad? • Simple • First-class Consul integration • Multiple workloads (Java, exec, Docker, ….)
28. node1 Nomad Consul & Nomad cluster node2 service-bar
 node1 service-foo
 node3 service-qux
 service- qux
 service- fox
 public-node1
29. Demo time!
30. Recap • Why did we do this? High availability • What did we gain? • High availability • Location transparency • Zero-downtime / rolling-upgrades • Resilience against machine failures
31. Recap Gradual migration
32. Thank you! @rkettelerij